GitLab

add documentation and types (+small fixes)

improve the example for scanette

based on a provided model of the system in dot format

update url of project
remove unused parameter in with_traces_split
update an Exception message

Remove fastcluster package as a dependency, since it is not used by agilkia, and causes install errors due to different numpy versions.

Add some simple tests to check that the examples/scanner scripts run with exit status 0.  This is helpful to catch basic API and syntax errors.

Change TraceSet.clusters to be private (._clusters) so it is not saved.  This is for consistency, since we cannot easily save _cluster_data and both are needed for visualize_clusters.  The idea is that clustering is transient, so needs to be redone after save/load.

Extend TraceSet.visualize_clusters to support separate markers for each cluster, to make scatter plots more readable.  It also generates a legend now, when there are multiple markers.

Added first version of SmartSequenceGenerator.generate_all_traces algorithm, to generate all high-probability traces up to a certain length.

Changed to English abbreviations for actions, for the paper.  Run all ML classifiers to measure their F1 scores.

Implemented TracePrefixExtractor (an sklearn feature extractor for trace prefixes) and SmartSequenceGenerator

Improve the default copying of meta_data when a TraceSet is created with traces all from one parent.

Add consistent meta_data field to all three levels: Event, Trace, TraceSet.  Make clustering and visualisation a bit more flexible.

Because conda-build cannot find it.